MLB Betting Model
I chose baseball because I knew nothing about it.
In 2003 I was losing money betting soccer.
I had played it since I was a boy. My father was a goalkeeper. It was the sport I loved.
That turned out to be part of the problem.
I had an opinion about everything.
So I went looking for a sport where I had no emotional attachment at all.
I found baseball.
And I started building.
Today the MLB Model creates a probability and fair price for every game, prices the first five innings separately and gives me an independent number to compare with the sportsbook market.
It shows me where my model price differs from the market.
Betting since 1998.
Building models since the early 2000s.
One-time payment · Lifetime access · 30-day money-back guarantee
The board
A number for every game.
A bet on only some of them.
The sheet gives me a number for the whole board.
That does not mean the whole board is actionable.
Most games should come back close enough to the market that I leave them alone.
That is useful information too.
A model is not there to create more bets.
It is there to give you something independent to compare with the price being offered.
A look inside
See the sheet before you buy it.
No black box.
This is the actual MLB model and the actual workflow.
Where this came from
2003.
Two options, and one of them was quitting.
By 2003 some of the edges I had relied on were disappearing.
I had been arbitraging and comparing prices between bookmakers, then betting soft local numbers when they differed enough from sharper markets.
It worked until access became the problem.
Meanwhile, I was betting soccer and losing.
I played the game.
My father played goalkeeper.
I knew the teams.
I had opinions.
And those opinions were not helping me.
So it was either stop or build something of my own.
Why baseball
I did not choose baseball because I loved it.
I chose it because it fit what I was trying to build.
A large betting market.
Huge amounts of statistical data.
And, most importantly, no emotional attachment from me.
I did not know the teams or players well enough to argue with the numbers.
I learned the sport.
Then I built the model.
Later I started publishing the work.
People began asking for the spreadsheet.
Marjan Balasko · Underdog Chance
Betting since 1998 · Building models since the early 2000s
The actual change
What is different the day after you buy this.
Without a number of your own, a price like:
−130
is just a number on a screen.
You can like the pitcher.
You can dislike the matchup.
You can know a team is hot.
But none of that tells you whether −130 is cheap, fair or expensive.
You need another price to compare it with.
After this, the morning looks different.
Open the sheet.
Confirm the starters and lineups.
Create your model probability and fair price.
Then compare it with the sportsbook.
Some days the answer is:
PASS.
Other days there is a difference large enough to investigate.
Either way, you now have a reason.
Inside the model
What it calculates.
Fair probability + fair odds
A probability for each game converted into decimal and American fair odds.
This is the independent price everything else starts from.
Market difference
Compare the model’s fair price with the sportsbook price.
The greater the difference, the more selective you can choose to become.
Projected score
The model also produces a projected score.
This can be useful for exploring totals, although full-game moneylines are the main market behind my historical record.
First five innings
The first five innings are priced separately.
This is not simply the full-game result divided up.
It places more emphasis on the starter and lineup and removes the later bullpen portion of the game.
In the season where I tracked full-game and first-five results separately, the full-game side made money and first-five lost money.
So I consider first five a useful additional model to work with, not the main reason to buy this product.
Underlying metrics
- —wRC+
- —recent wRC+
- —performance versus left/right-handed pitching
- —SIERA
- —xFIP
- —FIP
- —xERA
- —starter and bullpen data
- —recent / seasonal / longer-term weighting
If some of those names are unfamiliar, that is fine.
They are already built into the model.
You do not need to calculate them yourself.
The settings
Change the assumptions without rebuilding the model.
Most betting tools give you an answer.
This one also gives you the dials.
You can change how much weight goes to:
- —recent starting-pitcher form
- —season-long pitcher performance
- —longer-term performance
- —bullpen data
- —current lineup
- —team hitting
- —recent hitting
- —left/right-handed splits
Each weighting block must still add to 1.
The underlying calculation engine stays intact.
And the original settings remain visible beside your current ones, so you can experiment and still see where you started.
There is also a selectivity setting.
Higher means fewer bets qualify.
Lower means more.
My default is 1.5.
The work
The stats come in automatically.
The main version pulls the required data into Google Sheets automatically.
Your normal routine is:
- 01Open the model and let the data refresh
- 02Confirm the starting pitchers
- 03Confirm the lineups
- 04Enter the sportsbook prices you are comparing
- 05Read the model prices and differences
- 06Decide whether anything is worth betting
Websites change.
Scraping can break.
So you also get a manual copy-and-paste version of the model and a separate walkthrough showing how to use it.
Automatic when it works.
Manual fallback when it does not.
The training includes three videos:
- —how to use the MLB Model
- —how to update the model for future seasons
- —how to use the manual version
Approximately 50 minutes total.
Real bets
Ten tracked MLB seasons.
| Season | Units | Beat the line |
|---|---|---|
| 2016 | +40.67u | — |
| 2017 | +56.20u | — |
| 2018 | +3.83u | 64.00% |
| 2019 | +12.50u | 63.12% |
| 2020 | +15.20u | 77.78% |
| 2021 | −30.00u | 72.68% |
| 2022 | +23.44u | 74.35% |
| 2023 | +37.59u | 72.88% |
| 2024 | −21.19u | 73.89% |
| 2025 | −0.66u | 70.14% |
| Total | +137.58u |
Three of those ten seasons did not make money.
That is important.
A record that only shows winning seasons is advertising, not a record.
Historical results are not a guarantee of future performance.
Winning the bet and getting the right price are not the same thing.
Suppose I bet at decimal odds of 1.90.
The market later closes at 1.85.
That bet can still lose.
But I took a better price than the market eventually settled on.
That is what I mean by beating the line.
Across the seasons where I tracked it, roughly seven out of ten bets closed at a worse price than the one I had taken.
2021 lost 30 units while Beat the Line was 72.68%.
2024 lost 21.19 units while Beat the Line was 73.89%.
Those results are consistent with the reality that good pricing and short-term results do not always move together.
That does not make losing irrelevant.
It gives me another signal to inspect when deciding whether the process is still finding prices the market later agrees with.
One season, two markets
The full game made money.
First five gave it back.
Same spreadsheet family.
Two markets.
Very different results.
That is why I keep saying the goal is not to bet everything a model can price.
Find where your number is strongest.
Be harder to convince everywhere else.
There were also a handful of games where I entered the wrong starting pitcher and bet the value produced by that mistake.
I can calculate what the record would have looked like without those entries.
But I do not replace the published record with a corrected version afterwards.
The table stays −0.66u.
The practical lesson is worth more:
Check the starter.
Check the lineup.
And when the model suddenly shows a gigantic edge, check your inputs before you celebrate.
Be more selective
What happened when I raised the threshold.
55 higher-threshold bets produced almost as much profit as the full 345-bet sample.
That is one of the clearest examples from my own record of why I prefer selectivity to action.
In this 2023 sample, the more selective group also produced much stronger closing-price results.
Public proof
A public comment on the 2023 results post.
What people built with it
Brandon took the framework somewhere I never did.
Brandon was working in baseball in Asia.
He wanted to create numbers for Korean and Japanese baseball.
He did not ask me for picks.
He learned the framework and adapted the model ideas to KBO and NPB.
That is one reason I care more about teaching someone to build a number than sending them one of mine.
The limits
Read this before you buy.
Not a picks service
The model creates probabilities, fair prices and market comparisons. You decide what to bet.
Losing seasons exist
Three of the ten tracked seasons lost money. Losing seasons can happen again.
Full game is the strongest historical evidence
The model also calculates first five and projected scores. My historical record is strongest on full-game moneylines.
Check the inputs
Starting pitchers and lineups matter. Wrong inputs create wrong outputs.
Google Sheets
This is a spreadsheet built primarily for laptop/desktop use.
Baseball is daily
There are a lot of slates. The daily process is small, but it repeats throughout the season.
No guaranteed edge
Historical performance, CLV and beat-the-line results do not guarantee future profits.
Fit
Who this is actually for.
This is for you if
- —You want a baseball price of your own before looking at the sportsbook
- —You are willing to check starters and lineups
- —You like working with numbers, even a little
- —You prefer passing to manufacturing action
- —You want a finished model you can inspect and modify
Skip this if
- —You want picks sent to you
- —You will not open Google Sheets
- —You will not verify starting pitchers and lineups
- —You need a guaranteed profitable season
- —You want a completely automatic black box that requires no judgment
What you get
The MLB Model.
The automatic MLB Model
Google Sheets model with automatic data pulls.
The manual MLB Model
Copy-and-paste fallback version for when a source changes or automatic scraping is unavailable.
Full game pricing
Probabilities and fair odds.
First five pricing
A separately calculated first-five model.
Projected score
Additional output for exploring totals.
Settings page
Editable model assumptions and selectivity controls.
Three training videos
How to use the model, how to update for future seasons and how to use the manual version. Approximately 50 minutes total.
American odds bet tracker
Track results and closing-price information.
Decimal odds bet tracker
Track results and closing-price information.
Lifetime access
One payment.
No annual model subscription.
You keep the spreadsheet and training permanently.
The included training shows you how to update the model for future seasons.
Small bonus
50 complimentary AI Assistant credits.
The Underdog Chance AI Assistant is a separate product. The 50 credits let you try it on real games. Buying the MLB Model does not start an AI subscription. Nothing renews automatically. Credits are added manually after purchase.
Price
$499
One-time payment.
Lifetime access.
No recurring model fee.
30-day money-back guarantee
This is the most expensive individual model I sell.
It is also the one I have worked on the longest.
Two spreadsheet versions.
Full-game and first-five pricing.
Automatic data with a manual fallback.
And a model that has been part of my own betting process for more than two decades.
Buying several models? All Models is $997 and includes the complete model library plus A Journey training.
See all models →Delivery
What happens after you pay.
Your purchase is attached to the email address you use at checkout.
Go to My Library.
Enter that same email.
We send you a magic sign-in link.
Click it and your MLB Model content unlocks.
Inside you will find:
- —automatic spreadsheet
- —manual spreadsheet
- —training videos
- —included resources
Copy the Google Sheets files into your own Google Drive.
No software to install.
No password to remember.
Support: mb@underdogchance.com
Guarantee
30 days to use the real model.
Open it.
Run real slates.
Change the settings if you want.
Compare your model numbers with the market.
If you do not see the value, email me within 30 days and I will refund the purchase.
No call.
No proof required.
In their words
From people who used the work.
Member feedback
His reported MLB experience
Chef Alex’s public comment and my 2023 results post are shown in full further up this page.
Terry
On the videos and the numbers behind the odds
Authority
Comments from people with real experience in sports betting and statistics.
Dave Cokin
Dave Cokin is a longtime sports handicapper and respected voice in sports betting.
“Definitely worth reading for anyone trying to make their own numbers.”
David Beaudoin · Professor MJ
David Beaudoin, known as Professor MJ, is a PhD in statistics and professor at Laval University in Quebec City.
“His vision of sports betting is exactly like mine: you absolutely need to mix statistical models with your own judgement in order to be successful in this business.”
Questions
Before you decide.
How does the MLB Model work?+
The main spreadsheet pulls the required statistics automatically. You confirm the starting pitchers and lineups, enter the sportsbook prices you want to compare, and the model produces probabilities, fair odds and a market-difference value for the full game and first five innings.
Do I need to know SIERA, xFIP or wRC+?+
No. The calculations and initial weighting are already built into the model. The settings page explains the inputs if you want to understand or change them.
Do I need to be good at math?+
No. The formulas are already built. You need to be comfortable opening Google Sheets, confirming inputs and reading the output.
How long does a slate take?+
Usually around fifteen minutes once you know the workflow, depending on the number of games and lineup changes.
Do I need paid data?+
No. The model is designed around freely available sources.
What happens if the automatic data pull breaks?+
You also receive a manual copy-and-paste version and a video showing how to use it. Websites change sometimes, so the manual spreadsheet is the fallback.
What are first-five-inning bets?+
They settle using only the first five innings. The MLB Model prices that market separately with more emphasis on the starting pitchers and lineups.
Is the first-five model as proven as the full-game model?+
No. My historical record is strongest on full-game moneylines. In the season where I tracked the two markets separately, full game made +11.69 units while first five lost 12.35 units.
Is this a picks service?+
No. It creates a model price. You decide whether the sportsbook price is worth taking.
Which sportsbook do I need?+
Any sportsbook can be compared with the model. Access to multiple books can give you more opportunities to shop for a better price.
Does it work outside the USA?+
Yes. It is a Google Sheets model and I built and used it from Europe.
Will it work next season?+
You keep the spreadsheet and training permanently. One of the included videos shows how to update the model for future seasons. There is no annual model renewal.
Do you personally update my model every year?+
No. You receive the spreadsheet and training so you can update and use it yourself.
What does Beat the Line mean?+
If I bet at 1.90 and the market later closes at 1.85, I took the better price. The bet can still win or lose. Beat the Line tracks whether the price I took was better than the closing price.
Are the historical results guaranteed?+
No. The ten-season record describes my historical results. Three of those ten seasons lost money. Future results can be different.
Can I change the model?+
Yes. The settings page exposes the main weights and keeps the initial settings visible beside your changes.
Can I buy all the models?+
Yes. All Models is $997 and includes the complete model library plus A Journey training.
What if I do not like the MLB Model?+
Email me within 30 days of purchase and I will refund it.
I still have a question.+
mb@underdogchance.com
Last thing
The sport I picked on purpose.
I did not fall in love with baseball and then try to find a way to bet it.
I went looking for a sport where numbers could do the talking.
Baseball became that sport.
More than two decades later, this is still the model I have spent the most time building.
You do not need to spend two decades building yours.
You can start with the sheet I use, understand how it works, change the assumptions if you want and create a price of your own for the next slate.
One-time payment · Lifetime access · 30-day money-back guarantee
Stop following picks.
Start pricing bets.